A hyper-sphere SVM introduced the margin

Author

Xinfeng, Zhang ; Li, Zhuo ; Feng, David Dagan

Author_Institution

Signal & Inf. Process. Lab., Beijing Univ. of Technol., Beijing

fYear

2008

fDate

7-11 June 2008

Firstpage

470

Lastpage

475

Abstract

Binary hyper-sphere support vector machine (SVM) is a new method for data description. Its weakness is that the margin between two classes of samples is zero or an uncertain value, which affects the classifier´s generalization performance to some extent. So a generalized hyper-sphere SVM (GHSSVM) is provided in this paper. By introducing the parameter n and b (n>b), the margin which is greater than zero may be obtained. The experimental results show the proposed classifier may have better generalization performance and the less experimental risk than the hyper-sphere SVM in the references.

Keywords

pattern classification; sampling methods; support vector machines; binary hyper-sphere support vector machine; data description; generalized hyper-sphere SVM; pattern classification; sampling method; Biomedical signal processing; Equations; Face detection; Information processing; Medical diagnostic imaging; Neural networks; Signal processing; Support vector machine classification; Support vector machines; Generalization performance; Margin; hyper-sphere SVM;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks and Signal Processing, 2008 International Conference on

Conference_Location

Nanjing

Print_ISBN

978-1-4244-2310-1

Electronic_ISBN

978-1-4244-2311-8

Type

conf

DOI

10.1109/ICNNSP.2008.4590395

Filename

4590395